MPP 2019 - 8th Workshop on Parallel Programming Models - Special Edition on IoT and Machine Learning

Recent trends in artificial neural networks, such as deep neural networks, and the Internet-of-Things – IoT, indicate that an increasing number of artificial intelligence -based applications will be running on smartphones, sensors and other IoT devices collecting and processing large amounts of data. Most of those devices have limited processing power and often rely on cloud services for compute-intensive tasks. However, real-time applications may not tolerate the latency of offloading tasks to a cloud server.
Another important aspect to consider, especially in applications that run on big systems and manipulate big data sets, is the trade-off between moving data to a remote processing element to increase parallelism and computing things locally to reduce communication and energy costs while keeping performance levels. Edge/Fog computing proposes bringing computation closer to where data is sitting, by adding computational capabilities to network devices and adding edge gateways/servers, possibly in multiple layers with different latencies and computing performance. Moreover, such systems are expected to be heterogeneous, including multi-core processors, GPUs, FPGAs, and even processors that are customized for certain applications.
In this scenario, writing parallel applications is a non-trivial task, but also mandatory if one wants to explore the potential of the aforementioned modern computing platforms, imposing new challenges to the scientific community: the creation of models and alternatives to ease parallelism exploitation by the average programmer, considering the peculiarities of the different computation devices. Moreover, the proposed solutions should tackle problems such as application deployment, resilience and scheduling/offloading of tasks, considering latency, bandwidth, response time and computing power. In these complex environments, Machine Learning is becoming an important trend for the autonomic operation.
MPP aims at bringing together researchers interested in presenting contributions to the evolution of existing models or in proposing novel ones, considering the trends on IoT and Machine Learning. MPP 2019 will be held in conjunction with The 33rd IEEE International Parallel and Distributed Processing Symposium (IPDPS 2019), in Rio de Janeiro, Brazil on May 20-24, 2019.
Submission Guidelines
MPP invites authors to submit unpublished full and short papers on the subjects. Submissions must be in English, 8 pages maximum for full papers and 4 pages for short papers, following the IEEE formatting guidelines. Page limits includes references.
List of Topics
Topics of interest include (with special emphasis on IoT, Fog, Edge Computing, and Machine Learning) :
Novel execution models and languages for parallelism;
Novel parallel programming techniques and architectures;
Heterogeneous programming models;
Synchronization mechanisms;
Storage techniques;
Load-balancing and scheduling mechanisms;
Error detection/recovery;
Theoretical analysis of systems;
Smart network devices;
Software-defined networks;
Integration of IoT, Fog, Edge and Cloud Computing;
Neural Networks inference and training on IoT, Fog, Edge and cloud environments;
Performance analysis; and
Applications.